An approximate method for sampling correlated random variables from partially-specified distributions
Article Abstract:
An algorithm was designed for providing correlated vectors of random numbers where the user need not specify the total joint distribution function but only the marginal distributions and the correlation matrix. The algorithm may be applied for any set of continuous, increasing distribution functions which need not have similar functional forms. A procedure for adjusting the correlation matrix if the user-given matrix is not positive semi-definite was also provided. Two variations for the sampling algorithm were shown. The first tried to match the empirical correlation matrix as close as possible to the target correlation matrix to attain convergence in 100 samples. The second used a constant sampling variability in the empirical correlation matrix to attain convergence to the target correlation matrix only in very large samples.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1998
User Contributions:
Comment about this article or add new information about this topic:
An approximate method for sampling correlated random variables from partially-specified distributions
Article Abstract:
An algorithm was developed for producing correlated vectors of random numbers where the user need not provide the total joint distribution function but only the marginal distributions and the correlation matrix. The algorithm may be used for any set of continuous, increasing distribution functions which need not have a common functional form. A process for adjusting the correlation matrix if the user-given matrix is not positive semi-definite was also given. Two variations for the sampling algorithm were shown. The first tried to match the empirical correlation matrix as close as possible to the target correlation matrix to attain convergence in 100 samples. The second used a constant sampling variability in the empirical correlation matrix to attain convergence to the target correlation matrix only in very large samples.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1998
User Contributions:
Comment about this article or add new information about this topic:
A note on strategic sampling in agencies
Article Abstract:
The issue of statistical sampling for process control is examined with particular focus on the deterrence rather than the ex post detection of abnormalities. Deterrence takes precedence over ex post detection in strategic settings where the behavior of human agents can have considerable influence on the stochastic process. The study investigates sample design in a principal-agent setting where the performance of the agent who has control over the defect rate in a production process is measured and rewarded by sample outcomes. Calculation of the magnitude of gains from addition sampling is demonstrated, while the trade-off between these gains and sampling costs is examined. Findings show that the optimal sample size grows small as target defect rates are reduced.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1996
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: A new methodology for aggregating tables: summarizing journal quality data. The temporal nature of growth determinants in new bank foundings: implications for new venture research design
- Abstracts: Lucent says profit for latest quarter will more than double projections. Cabletron CEO is expected to resign
- Abstracts: Modelling and forecasting UK public finances. Minimum wages: possible effects on the distribution of income
- Abstracts: Fun and games on a grand scale. Not really retired, just out to pasture. Master of ceremonies
- Abstracts: New 'indirect' valuation method likely to have long-term effects